The web UI is super slow. It's slow with the default view of 10 results per page and unbearably slower at 100. These aren't big numbers and my computer isn't old, something was poorly done here and setting the default number of results per page to 10 clearly is not the correct fix.
(For the record it only has a mere thousand results so far..)
Also, the content classification is so bad it might as well not exist. How does a torrent with "Playboy" in the title get classified as 'Unknown' instead of 'XXX'? Even torrents with "Porn" or "XXX" in the title get classified as Unknown. A simple Bayesian classifier should have this covered, it's not a task that needs heavy duty AI to solve.
Try https://www.coveapp.info/, I mentioned it here: https://news.ycombinator.com/item?id=39425381. I wrote a custom search implementation for torrents that uses keyword matching and boosts search scores based on popularity of torrents on the DHT. It feels very much like TPB search but not so aggressively focused on seeder counts (i.e. better matching on file names can outrank a higher seeder count).
Can I take a wild guess that you're using Firefox? I have noticed Firefox performance is much worse than any other browser. I think this is due to issues in the Angular and Material components being used.
As stated in the big red notice in the website, the software is currently in alpha preview. Given time, Bitmagnet will be trying to mitigate the Firefox performance issues (the <5% market share means it's prioritised accordingly), but it's important that users of all browsers can have acceptable experience in the app.
It takes several seconds to load 100 results from a list of 1000, with no search term used. I don't know how that's even possible. I think half-baked is a mild way of putting it, particularly if the excuse is that it hasn't been in the oven very long.
(For the record it only has a mere thousand results so far..)
Also, the content classification is so bad it might as well not exist. How does a torrent with "Playboy" in the title get classified as 'Unknown' instead of 'XXX'? Even torrents with "Porn" or "XXX" in the title get classified as Unknown. A simple Bayesian classifier should have this covered, it's not a task that needs heavy duty AI to solve.
This whole thing seems way too half-baked for me.